Quantifying the Effects of Database Size and Sample Quality on Measures of Individualization Validity and Accuracy in Forensics

The researchers propose to develop methods to statistically quantify (1) the uncertainty in measures aimed at validating a forensic discipline's basic premises (such as a uniqueness claim) and (2) the use of likelihood ratio methods in making classification/individualization conclusions. The use of automated pairwise comparisons of biometric samples in a database is a basic element of forensic individualization determinations involving biometrics, such as fingerprints and handwriting. An issue that applies to forensic individualization is that while a database of samples can be used to support individuality, it does not necessarily prove individuality. Phase I of this proposed project will focus on the random match probability (RMP), as a measure of the validity of a forensic individualization procedure. In Phases II and III, the researchers will shift focus to quantifying accuracy with likelihood ratio methods. The researchers propose to investigate its estimation in other fields outside of DNA, such as handwriting and glass fragments, focusing both on statistically sound point estimates and confidence intervals. Most of the methodologies developed in this proposed project will apply to any field of forensics as RMPs and likelihood ratios are defined similarly in many of them. ca/ncf